In view of the problem that the alignment of internal and external detection data mainly relies on manual verification and excavation verification, and can not make full use of mining detection information, feature information and mileage information are extracted from internal and external detection data as input and output variables, and the mapping relationship between them is established by using limit gradient lift (XGBoost) algorithm. Predict the outer detection mileage of the internal detection point, take the internal detection information as the benchmark, and use step by step translation to realize the alignment of the internal and external detection data, and conduct a comprehensive analysis according to the alignment results. The results show that the mileage error of all stations and valve chambers after data alignment is within 5m, the latitude and longitude distribution changes in the same height, and the alignment result points meet the accuracy requirements. In the example analysis, the degree of AC and DC interference of the pipeline is small, and the cathodic protection of the pipeline is normal. Although the soil corrosion is strong, there is metal loss and no obvious corrosion pits and corrosion products are found at the damage of the corrosion layer, indicating that the pipeline is in good running condition. The research results can provide theoretical basis for improving the level of pipeline integrity management.